Rainfall,groundwater and refuges: Predicting extinctions of Australian tropical mammal species

Abstract A general decline in populations of some savanna mammal species has occurred since the mid-1980s in a fairly pristine national park environment in the Australian wet-dry tropics. Terrestrial native rodents have exhibited the greatest decline and marsupials the least. During the same period, waterholes have dried up and the previous strong association of mammals with riparian vegetation has diminished. A regional index of the level of groundwater was developed from 16 unexploited bores from across the northern half of the Northern Territory. Predicted mean groundwater values over 28 years showed a strong increase throughout the 1970s and a similar decrease throughout the 1980s. Using data since 1986, strong correlations between mammal numbers and groundwater levels for the preceding two years, and much weaker relationships with rainfall, were obtained. In turn, the groundwater levels were best correlated with an eight-year running mean of rainfall for both Darwin and Alice Springs and an eleven-year running mean for the Southern Oscillation Index. Extrapolating backward in time, cumulative pressure variation at Darwin showed two extended periods of predicted low ground-water values, 1900s–1920s and 1940s. Both periods were preceded by the last records of a number of now-extinct mammals from central Australia. A third trough in the early 1960s suggests another period of extinction in central Australia, previously unappreciated due to the lack of survey work in the 1950s. Further, by this analysis, the past 20 years has been the best period for mammals since weather records commenced. Analysis of continental-scale distributions of mammals showed evidence of fluctuations suggesting rocky areas are important refuges for some species during periods of low groundwater levels. In contrast to those in savanna habitats, the mammals of the extensive wetland areas fluctuated in harmony with rainfall on a short-term basis except where habitat is flooded. It is the species which occur only in savanna but not also in rocky or wetland habitats which are most at risk. A number of other species-vulnerability characteristics were identified: riparian vegetation specificity; the degree to which they fall within the critical size range (35 g-5. 5 kg); degree of semelparity; smallness of population size; and smallness of geographic range. Those savanna species with annual life histories strongly associated with riparian areas are likely to be in greatest jeopardy. The species which appear to be most vulnerable in northwestern Australia are Conilurus penicillatus, Mesembriomys gouldii, Mesembriomys macrurus, Antechinus bellus, Phascogale tapoatafa and Rattus tunneyi. Our recent benign climatic history has coincided with the great public environmental awakening and upsurge in environmental research and management efforts. This historical accident has probably led us to overstate the negative effects of human impact and also our ability to change the course of biological history by ameliorating human impact. Our distorted view of these factors has further led us to over-emphasize research of a spatial nature to the detriment of temporal studies. We must redress this balance.